Frequently Asked Air Questions (FAAQs): Understanding Data Center Emergency Generators and Their Air Permits

Summary – Emergency generators maintain data center operations during power outages, but their growing use has also raised concerns about air quality, energy demand and environmental impacts. The article provides an overview of how emergency generators are regulated under federal and state air permitting programs, how they are typically permitted at data centers through synthetic minor permits and the operational and compliance considerations associated with those permits.

Emergency generators are an essential tool because they allow businesses to continue to operate—without a gap in operations, loss of data, or any other interruption—during a sudden loss of power. Moreover, the need for emergency generators is ever-growing, as industries like data centers desire the promise that access to emergency power provides during an outage: the preservation of the integrity of their data, continued customer access to hosted systems, and the functioning of key infrastructure, such as cooling systems. Nevertheless, this increased demand for emergency generators has "energized" debate over the last year as concerns grow about data centers, energy costs, where data centers are sited, and their potential impacts to the environment—including to air quality. For example, it is estimated that Virginia has permitted over 9,000 diesel emergency generators for back-up power at data centers to date, but the general public may lack overall background knowledge about these air permits. At the same time, emerging data centers and related industries grapple with formulating the operational terms of their air permits and then maintaining compliance with those terms.

The aim of this Frequently Asked Air Questions (FAAQs) article is to provide an overview of the interplay between emergency generators and air permitting and illustrate how emergency generators are typically permitted at data centers. This article should ultimately provide valuable background to both affected community- and industry-members alike on a topic that is "generating" more and more attention as data centers expand across the country.

Note: At the federal level, the key location for identifying what distinguishes emergency generators from non-emergency generators are the regulations set forth in the New Source Performance Standards (NSPS) Subparts IIII (Stationary Compression Ignition Internal Combustion Engines) and JJJJ (Standards of Performance for Stationary Spark Ignition Internal Combustion Engines), and National Emissions Standards for Hazardous Air Pollutants (NESHAP) Subpart ZZZZ (Reciprocating Internal Combustion Engines ("RICE")). The NSPS provisions effectively split the "universe" of engines based on how an engine ignites its fuel, whereas the NESHAP does not distinguish engines on this basis. For non-lawyer or non-technical readers, in this article the terms engine or generator are synonymous.

Q: Generally, how do generators affect air quality?

A: Generators emit criteria pollutants, i.e., pollutants regulated under the Clean Air Act's National Ambient Air Quality Standards (NAAQS), as fuel is combusted. In particular, generators have the potential to emit nitrogen oxides (NOx), carbon monoxide (CO), particulate matter (PM), and volatile organic compounds (VOCs). Additionally, generators also have the potential to emit hazardous air pollutants (HAPs), such as formaldehyde, benzene, and other combustion byproducts, depending on the fuel used and engine design.

The concentration of pollutants emitted from a generator set may decrease or increase depending on what type of fuel the generators are utilizing. Fuel selection is one area where there may be an opportunity for data centers to potentially reduce their emissions. For example, a Rolls Royce study on its mtu brand of generators identified that the use of hydrotreated vegetable oil (HVO) resulted in less NOx and PM emissions, as compared to the use of distillate diesel fuel. In particular, Rolls Royce determined that a NOx reduction of 8% occurred with HVO as compared to diesel, with a higher reduction observed at lower loads; and a reduction of PM emissions from HVO as compared to diesel ranged from 50-80% depending on load point. Consequently, data centers may have the opportunity to decrease their impacts to air quality based on the type of fuel employed at their facility and the type of generators installed.

Critically, generators permitted as "emergency" generators are less likely to cause negative air impacts, as emissions are intended to be both intermittent and limited in duration. 

Q: What is an emergency generator?

A: An emergency generator is a generator that is "operated to provide electrical power or mechanical work during an emergency situation."[1] Generally, emergency generators have restrictions on their use (i.e., use during an emergency and certain non-emergencies only), and are subject to different regulatory requirements than non-emergency generators—such as notification requirements, emissions standards, monitoring standards, and recordkeeping requirements.

Q: What is (and is not) an "emergency" for purposes of operating an emergency generator?

A: Because emergency generators are generators intended to be operated during "emergency situations," it is important to understand what constitutes an "emergency" under the federal rules. While EPA does not define "emergency situation," it provides two examples of what constitutes emergency use: (1) when a loss of power from a local utility or normal power source occurs; or (2) if a fire or flood occurs requiring use of an emergency generator to pump water.[2]

Because the Clean Air Act effectively operates as a regulatory floor at the federal level, it is important to note that states may adopt their own definition of an "emergency" or "emergency situation" that may be as strict or more strict concerning what constitutes an “emergency.” For example, in Virginia "emergency" means a "condition that arises from sudden and reasonably unforeseeable events where the primary energy or power source is disrupted or disconnected due to conditions beyond the control of an owner or operator of a facility including: a failure of the electrical grid, on-site disaster or equipment failure and public service emergencies such as flood, fire, natural disaster, or severe weather conditions.[3] North Carolina’s definition, by contrast, does not expressly hinge on the event being sudden or unforeseeable, but instead focuses on whether the loss of primary power was beyond the owner or operator’s control: “the loss of primary power at the facility that is beyond the control of the owner or operator of the facility.”[4]

Q: Can emergency generators ever run during non-emergencies?

A: Generally, yes. Under the federal rules, emergency generators can run for a limited amount of time during "maintenance and testing hours" and "non-emergency situation hours."

Maintenance and Testing Hours: The federal rules state that emergency generators can be operated for "maintenance checks and readiness testing, provided that the tests are recommended by federal, state or local government, the manufacturer, the vendor, the regional transmission organization or equivalent balancing authority and transmission operator, or the insurance company associated with the engine."[5] EPA has stated that maintenance and testing hours can include training for and simulation of emergency situations, as well as documented engine repair and associated confirmatory testing.[6] 

Non-Emergency Situation Hours: The federal rules do not explicitly define other "non-emergency situations." However, the federal rules provide examples of allowed non-emergency situations and prohibited non-emergency situations. 

Allowed: Emergency generators are allowed to be used to supply power as part of a financial arrangement with another entity only if certain conditions are met. These conditions include: (1) the engine is dispatched by the local balancing authority or local transmission and distribution system operator; (2) the dispatch is intended to mitigate local transmission and/or distribution limitations so as to avert potential voltage collapse or line overloads that could lead to the interruption of power supply in a local area or region; (3) the power is provided only to the facility itself or to support the local transmission and distribution system; and (4) the dispatch follows reliability, emergency operation or similar protocols.[7] Another example of permissible non-emergency use of an emergency generator is during a planned outage.[8] 

Prohibited: Emergency generators are not allowed to be used for peak shaving or non-emergency demand response, or to generate income for a facility to an electric grid.[9]

EPA appears to draw the line between permissible and prohibited non-emergency use based on whether the operation is fundamentally related to reliability and intended to help avert grid instability, as opposed to other forms of operation. For example, EPA has characterized peak shaving as being “designed to increase capacity in the system” and “generally based on the economic benefit gained by operating the engine rather than another power source.”[10]

Finally, it is important to note that, because states may adopt more stringent requirements under the Clean Air Act framework, states may prohibit the operation of emergency generators during non-emergency situations altogether. Virginia, for example, allows only non-emergency operation of “mission-critical” emergency generators.[11]

Q: What hour limits are emergency generators subject to under the federal rules?

A:

Importantly, EPA has suggested that in the event these hour limits are not followed, an emergency generator will no longer be considered an emergency generator under the NSPS and NESHAP standards. Consequently, the generator would need to meet all requirements for non-emergency generators at the time the generator changed its status from emergency to non-emergency, including any applicable emission limits, performance testing, and other requirements for a non-emergency generator. This could require replacement of the generator with a new, low-emissions generator, or adding major controls to the generator to meet applicable emission limits.[12]

Q: How are most emergency generators permitted?

A: Under the Clean Air Act, sources may be subject to “New Source Review” (NSR) permitting, which is a pre-construction permitting program that applies to new or modified sources of air emissions. The goal of NSR is to ensure that newly constructed or modified sources consider their air emissions and protect the air quality of the geographic area in which they are located prior to construction. Most emergency generators at large data centers are permitted under synthetic minor permits. Although individual generators may be relatively small, data centers often contain dozens of generators whose emissions must be aggregated for permitting purposes. This aggregation results in a potential to emit that could otherwise exceed major source thresholds under the Clean Air Act.

Q: What is a synthetic minor permit?

A: A synthetic minor permit is a permit that subjects a source to enforceable limits that will restrict its potential to emit air pollutants to a level below the threshold levels applicable to a major source. The namesake “synthetic minor” refers to the fact that the source has the capacity to be a major source of emissions but is choosing to operate “synthetically” as a minor source.[13]

Q: How can you tell if a permit is a synthetic minor permit?

A: A synthetic minor permit will often expressly state that the permit is intended to establish enforceable limits keeping the facility below major source thresholds under the Clean Air Act. For example, one Georgia permit for a data center with over 50 diesel emergency generators states: “This Permit is issued for the purpose of establishing practically enforceable emission limitations such that the facility will not be considered a major source with respect to Title V of the Clean Air Act Amendments of 1990.”

This language reflects the core purpose of a synthetic minor permit: the source could otherwise emit at major source levels, but agrees to enforceable limits to remain below those thresholds.

Q: What are the “enforceable limits” data centers use to limit their potential to emit?

A: Enforceable limits or enforceable restrictions are conditions in a permit that limit a source’s potential to emit air emissions. These conditions at data centers generally fall into two categories.

• Emission Limits: These are restrictions over a given period of time that limit the amount of pollutants that may be emitted from a source.
• Operation Limits: This is a catch-all category for all other restrictions on how a source will operate, including limits on hours of operation, amount of raw material consumed, or fuel allowed to be combusted. This category also includes conditions that outline what add-on controls must be installed and maintained at a source, and what its emission rate or efficiency is required to be.

Typically, emission limits and operational limits work in tandem. For example, a synthetic minor permit may contain a throughput limit restricting the amount of fuel that may be combusted by the data center’s generators over a specified period of time. The corresponding emission limits for the data center are then based on the quantity of air pollutants expected to be emitted from combusting that amount of fuel.

It is important to note that whatever limitations are incorporated into a data center's synthetic minor permit, those limitations must be both federally and practically enforceable. Federally enforceable essentially means that it is contained in a permit issued pursuant to an EPA-approved program or directly by the EPA or has been submitted to EPA as a revision to a State Implementation Plan. Practically enforceable relates to whether the EPA or a state agency can monitor compliance with those restrictions and determine if a source has been effectively operating as a synthetic minor source or as a major source. For example, if a permit limits the throughput for a source, in order for that limit to be practically enforceable, the permit should also include separate monitoring and recordkeeping requirements so that someone can verify the actual amount of fuel combusted at each generator.

Q: Are there any tradeoffs to using an hours limit vs. a throughput limit as an operational limit at a data center?

A: Each approach has advantages and disadvantages, especially for facilities that may contain a large number of emergency generators.

Hours limits are generally simpler to understand and administer because they restrict the number of hours a generator may operate over a specified time period. However, hours limits can be easier to exceed during prolonged outages, particularly where a facility operates many generators simultaneously. For example, if a data center has 50 permitted generators and a 3,000 hour limit, if the data center experiences an outage during extreme weather lasting more than two days, the generators could collectively consume all of their allowable annual operating hours in a single event—and that window becomes even narrower if the facility has 100 permitted generators instead. 

Hours limits also typically assume operation at a constant load associated with a conservative or worst-case emission rate. In practice, however, generators often operate at lower loads during outages, meaning actual emissions may be lower than the emissions assumed by the permit. As a result, a data center can exceed its allowable operating hours while still emitting less pollution than the permit’s modeled worst-case assumptions.

Throughput limits, in contrast, can better account for fluctuations in actual generator usage and load conditions during an outage, because generators consume less fuel at lower and/or different operating loads. However, throughput limits and their associated emission limits are still generally based on assumed emission factors tied to worst-case fuel combustion; consequently, the emission limits may not perfectly account for differences in emission rates at varying loads during an outage, either. 

Q: Are there ways to account for actual generator load and emissions more accurately in a permit?

A: Potentially, yes. Some approaches may allow a source to better account for actual generator emissions and avoid some of the tradeoffs associated with broad worst-case operational assumptions like an hours or throughput limit. For example, in some circumstances permittees may physically or electronically restrict generator loading through control systems or other operational limitations. Where such restrictions are practically enforceable and verifiable, permitting authorities may allow emission calculations and lower permitted maximum emission rates based on lower loads rather than worst-case-full-load assumptions. Such restrictions, however, may reduce operational flexibility if a data center later increases capacity or electrical demand and needs the generators to operate at higher output levels. Such a change would likely be considered a major permit modification.

Similarly, sources may consider stack testing generators at lower load bands to confirm manufacturer emission specifications at those operating conditions. This can provide a more representative basis for calculating emissions during an outage rather than assuming continuous operation at maximum load or the worst-case emission rate. EPA guidance recognizes that engine performance testing can reflect site-specific operating conditions and achievable load, further suggesting that generator load conditions may be relevant to how emissions are evaluated and demonstrated for compliance purposes.[14]

Q: What are some advantages to being a synthetic minor source?

A: Overall, the permitting process for a synthetic minor source is less demanding than the permitting process for a major source. Moreover, because a NSR permit is required prior to construction, a synthetic minor permit may allow a source to initiate construction sooner—as major source permits require more of an exacting review process, so it takes longer for the issuing agency to finalize the permit.

In addition, major sources are subject to more stringent control measures, so a source may desire to be a synthetic minor source to avoid the costs of complying with those measures. Major PSD NSR sources are required to include best available control technology (BACT) measures in their operations. BACT determinations balance energy, environmental, and economic impacts and these determinations can require a source to undergo equipment modifications, incorporate new combustion techniques, or comply with various operational standards, design standards, or work practices. These BACT determinations must be made for each regulated NSR pollutant that the source would have the potential to emit in "significant" amounts, and complying with BACT can be burdensome.

Q: Do synthetic minor permits undergo public comment?

A: EPA has stated as recently as 2022 that its SIP rules generally require permitting agencies to provide an opportunity for public comment for NSR permits, including synthetic minor permits establishing enforceable limits used to avoid major source permitting requirements.[15] EPA has also acknowledged, however, that permitting approaches may vary among agencies, including the use of different public participation thresholds or federally enforceable state operating permit (“FESOP”) programs for certain synthetic minor sources.[16] Additionally, in a 2021 EPA Inspector General Audit Report, the agency asserted that synthetic minor source programs in each state must allow the public to participate in the permitting process. The Report further recommended that EPA work with all state, local, and tribal agencies that do not provide the opportunity for public comment on synthetic minor source permits, to assure opportunities for public participation occur.[17]

Q: What other permit terms are worth reviewing closely in data center synthetic minor permits?

A: In addition to operational limits such as hours limits or fuel throughput limits, synthetic minor permits for emergency generators often contain a number of other important conditions that can significantly impact facility operations and compliance obligations.

For example, many permits require initial stack testing or performance testing shortly after the generators become operational. Because emergency generators typically cannot be stack tested prior to installation and startup, permitting authorities often initially rely on manufacturer specifications during permitting, while later requiring site-specific testing to confirm compliance with applicable emission limits under actual “real world” operating conditions.

Permits may also contain detailed notification and reporting requirements tied to installation or operation. Some permits may require prompt notification to the state agency following certain outage events, exceedances, or operation beyond specified emergency-use parameters.

In ozone nonattainment or maintenance areas, permits may also include seasonal restrictions or additional operational limitations during ozone season because diesel-fired generators can emit significant quantities of NOx, an ozone precursor. Permits can also restrict the timing and duration of testing activities to mitigate associated noise pollution, localized air quality impacts, and other community concerns.

[1] See e.g., NSPS Subpart IIII, at § 60.4219.
[2] See e.g., NSPS Subpart IIII, at § 60.4219.
[3] 9 VAC5-80-1110
[4] 15A NCAC 2D .1401(13)(A).
[5] See e.g., Subpart IIII, at § 60.4211(f)(2)(i).
[6] 71 Fed. Reg. 39154, 39165 (July 11, 2006).
[7] See e.g., Subpart IIII, at § 60.4211(f)(3)(i).
[8] See Region 7 Letter to Missouri Public Utility Alliance Re: 40 CFR Part 63 National Emission Standards for Hazardous Air Pollutants for Reciprocating Internal Combustion Engines; Final Rule Dated March 3, 2010, available online here.
[9] See e.g., Subpart IIII, at § 60.4211(f)(3).
[10] 78 Fed. Reg. 6674, 6686 (January 30, 2013).
[11] See 9VAC5-80-1111 (providing 50 non-emergency hours, including for planned outages, to emergency generators certified by the U.S. Department of Defense “as mission-critical and essential to the defense of the United States.”).
[12] EPA, Implementation Question and Answer Document for National Emission Standards for Hazardous Air Pollutants for Stationary Reciprocating Internal Combustion Engines and New Source Performance Standards for Stationary Compression Ignition and Spark Ignition Internal Combustion Engines (April 2, 2013), available online here.
[13] For more information about synthetic minor permitting, see Frequently Asked Air Questions (FAAQs): Synthetic Minor Permits, One-Time Doubling & Sham Permits, available online here.
[14] SeeEPA, Implementation Question and Answer Document for National Emission Standards for Hazardous Air Pollutants for Stationary Reciprocating Internal Combustion Engines and New Source Performance Standards for Stationary Compression Ignition and Spark Ignition Internal Combustion Engines (April 2, 2013).
[15] See EPA, Clarifying/Revising the Minor New Source Review Program for Air Agencies; Fall 2022 Stakeholder Engagement Sessions, available online here.
[16] Id.
[17] EPA Office of Inspector General, EPA Should Conduct More Oversight of Synthetic Minor Source Permitting to Assure Permits Adhere to EPA Guidance, Report No. 21-P-0175 (July 8, 2021), available online here (“Public participation requirements are intended to assure that citizens have the ability to provide input on the permit limitations, the pollutants that are emitted, and the quantity of emissions that will affect their communities. Not providing the opportunity to review permit materials and provide comments on proposed synthetic-minor source permits hinders citizens’ ability to weigh in on facilities that may impact air quality where they live.”)